Mail Archives: geda-user/2016/05/31/05:06:56
On Tue, May 31, 2016 at 09:43:03AM +0200, gedau AT igor2 DOT repo DOT hu wrote:
> Hi all,
>
> first of all: this mail has nothing to do with gschem and does _not_
> propose anything for mainline pcb either. I state generic "what
> users would prefer in a random software having the same issue"
> questions.
>
> (Background: I'm working on the "conf rewrite" in pcb-rnd,
> collecting all the configuration/setting/preferences data into a
> central tree in memory that has an uniform API.)
>
> Today I bumped into an old problem that does affect users who want
> to keep .pcb files in VCS: the Cursor line and the Grid line noise.
>
> The core of the problem is that these two lines represent GUI
> settings. On one hand it'd be great to have the last cursor position
> and grid setting saved but it certainly can interfere with VCS usage
> generating dummy diffs.
>
> With the conf rewrite I have a new perspecitve looking at this old
> problem and I came up with this idea:
>
> - a design file should not contain non-design-related data (e.g. GUI
> settings)
>
> - instead there could be a separate file that could save such settings
>
> - the user would place the main design file (e.g. foo.pcb) under VCS
> but the design-preferences file with all gui settings would be an
> uncommited local file (e.g. foo.pcb.pref).
>
> - drawback: the user needs to know about what the two files are for,
> when adding one (or both) to vcs, when sharing/distributing them,
> during backups, etc.
>
> So my question is, as an user, what do you think about this
> approach? Would it bother you if a random CAD program would start to
> split design files like that? Is it worth the hassle (from the
> user's point of view) or the Cursor-Grid-like-noise is too small to
> deal with 2 files all the time?
My position is that it is too small for these two lines. Actually I
always thought that keeping the cursor position was overkill.
This said, there are other flags, like "selected" and "found" which in
my opinion only add noise to the diffs, and should not be saved in the
main pcb file. Keeping them in an associated file might be fine, however.
Anothe solution would be to have a specialized diff program for the
PCB files. For people using git, I suspect that it would not be very
hard to use an external (GIT_EXTERNAL_DIFF) diff program that would
present actual differences and filter out what can be considered noise.
I don't know whether other VCS allow something similar.
In any case, it's only my opinion and you write the code, so you decide.
Regards,
Gabriel
- Raw text -